Sun induced fluorescence (SIF), the radiation flux emitted by plant chlorophylls molecules in the 650-800 nm spectral window, is considered an indicator of photosynthetic performance. Recently it has been shown that SIF can track changes in light use efficiency (LUE), and therefore it is a good predictor of gross primary production (GPP) at various scales, from leaves and ecosystem to regional and global scale. Although SIF has been successfully used to predict GPP in various ecosystems, the mechanistic link between GPP and SIF remains not fully understood, and especially the effect of function and structure on SIF at the canopy scale remains an active area of research. SIF is emitted by the whole canopy, but only a fraction of the total emission is observed with remote sensing techniques. The escape probability of SIF (Fesc) controls the amount of SIF scattered by the canopy and is integral to separate the effect of canopy structure and function on the fluorescence signal. In this contribution we make use of data collected at the research site Majadas del Tietar, a Mediterranean grassland manipulated with Nitrogen and Phosphorus. Using the SCOPE model (Soil Canopy Observation Photochemistry and Energy fluxes) we obtain Fesc and we analyse how Top of canopy SIF and emitted SIF vary in response to the fertilization. With a combination of processes-based modelling and data driven analysis, such as relative importance analysis and structural equation modelling, we unravel the processes and causal relationship that are at the base of the GPP - SIF relationship. We show that the nutrient fertilization had an effect on plant composition, and therefore canopy structure, but also plant functioning. Nitrogen induced changes in biodiversity mainly affect leaf angle distribution of the canopy and therefore scattering properties such as Fesc. The nitrogen fertilization is also responsible for a change in plant functioning, with altered SIF emission. The simultaneous change of both canopy and structure causes the fertilization effect to be visible mainly at the emission level, but not at top of canopy, as the variation in canopy structure masks the change observed at leaf level. This contribution advances the knowledge of the highly complex dynamics involved in the GPP-SIF relationship. In depth understanding of the mechanistic processes is required to fully take advantage of the increasingly prevalent SIF data streams.